State of Open Source on Hugging Face: Spring 2026

Hugging Face published its Spring 2026 analysis examining shifts in the open source AI landscape across competition, geography, technical trends, and emerging communities. The platform now hosts 13 million users, more than 2 million public models, and over 500,000 public datasets.

Activity has nearly doubled across users, models, and dataset repositories. Users increasingly create derivative artifacts such as fine-tuned models, adapters, benchmarks, and applications rather than only consuming pre-trained systems.

The ecosystem remains highly concentrated, with approximately half of all models having fewer than 200 downloads. The top 200 most downloaded models, representing just 0.01% of all models, comprise 49.6% of total downloads.

Over 30% of Fortune 500 companies maintain verified Hugging Face accounts. NVIDIA has emerged as the strongest Big Tech contributor, while Thinking Machines built its Tinker model entirely on open weights and IDEs like VSCode and Cursor support both open and closed models.

China has surpassed the U.S. in monthly and overall downloads, accounting for 41% of downloads in the past year. Models from individual users or distributed organizations without a clear geographic base represent about half of all platform downloads.

Industry's share of development fell from approximately 70% before 2022 to roughly 37% in 2025. Independent developers rose from 17% to 39% of downloads, with individuals and small collectives focused on quantizing, adapting, and redistributing base models.